SUMMARY Our overall goal is to systematically define c/s-regulatory elements in mRNAs that control translation initiation under defined physiological conditions. While the sequence requirements for translation start codon selection have been studied for individual genes, a comprehensive understanding of the rules for start site selection is not known for the human transcriptome. Our aims are to: A|m 1) Determine translation initiation site usage in human cells in defined physiological states. Aim 2) Map mRNA structure in vivo and in vitro using chemical probing methods. Aim 3) Determine the positioning of eukaryotic initiation factor elF3 on mRNAs in human cells. Aim 4) Develop computational models for predicting start codon selection. Our efforts will initially focus on developing heuristic models for how start codons are selected in human cells. These models will combine the RNA structural information obtained in the above aims with phylogenetic parameters to increase the models' predictive power. As we build a knowledge base for human mRNA structural properties, we will incorporate principles of RNA structure into these models. Our ultimate goal will be to couple general principles of RNA structure to systems-level experimental constraints to enable the prediction of translation initiation patterns in human cells, using minimal new inputs for a given cell type, tissue, or physiological state.